This is a very limited effect against a motivated hacker or AI red teamer, because a lot of these times they can exploit what I call the intelligence gap between these guardrails and the main model where say I Base64 encode my input. A lot of times the guardrail model won't even be intelligent enough to understand what that means.
This is a very limited effect against a motivated hacker or AI red teamer, because a lot of these times they can exploit what I call the intelligence gap between these guardrails and the main model where say I Base64 encode my input.